In July 2004, the Belgian government intensified monitoring within the Unemployment Insurance scheme. Workers claiming unemployment benefits for more than 13 months are notified that past job-search behavior will be monitored 8 months later. In one region the target group is counselled shortly after the notification, while in others not or only once the monitoring has taken place. We exploit the discontinuity in the treatment assignment at the age of 30 to evaluate the threat effect of the notification on the probability of employment. We find that the effect is heterogeneous and critically depends on whether and when notified workers are counselled.